CO₂驱是目前低渗透油藏提高采收率的重要手段，但受油藏非均质性影响，长期注气极易导致CO₂气窜，使得油藏中存在大量剩余油，极大影响CO₂驱开发效果。CO₂水气交替驱（CO₂ WAG）是一种抑制低渗油田CO₂气窜的有效技术，其实施过程中涉及注入速度、段塞大小和气水比等众多注入参数，不合理的注入参数难以发挥其提高原油采收率作用。传统油藏数值模拟方法确定最优注入参数方案费时费力，成本高，大型油田多井复杂注入参数组合下甚至难以实现。该研究将饥饿游戏搜索算法引入CO₂水气交替驱注入参数优化过程，并利用混沌映射函数提高其初始注入参数取值的随机性和多样性，形成一种新的混沌映射函数改进饥饿游戏搜索算法，实现算法与油藏数值模拟软件的协同智能优化，提高典型油田CO₂水气交替驱注入参数优化的精度和效率。研究表明：与Logistic、Gussia和Singer混沌映射函数相比，Tent混沌映射函数所得混沌值和频数分布更加均匀，适合于改进饥饿游戏搜索算法。Tent混沌映射函数改进饥饿游戏搜索算法是一种有效的CO₂水气交替驱注入参数优化方法。该算法所得CO₂水气交替驱最优注入参数方案累积产油量为34.974×10⁴ m³，比饥饿游戏搜索算法所得累积产油量增加0.213×10⁴ m³，比现有CO₂水气交替驱注入参数方案增加5.820×10⁴ m³，为现场CO₂水气交替驱高效实施提供了有效技术手段。

CO₂ flooding is an important method to enhance oil recovery in low-permeability reservoirs. However, due to the heterogeneity of the reservoir, long-term CO₂ injection can easily lead to CO₂ gas channeling, leaving a large amount of residual oil in the reservoir, which severely impacts the effectiveness of CO₂ flooding. CO₂-water-alternating-gas flooding (CO₂ WAG) is an effective technique to suppress CO₂ gas channeling in low-permeability oilfields. During the implementation of CO₂ WAG, numerous injection parameters such as injection rate, slug size, and gas-water ratio are involved. Unreasonable injection parameters make it difficult to achieve improved oil recovery. Traditional reservoir numerical simulation methods for determining optimal injection parameters are time-consuming, labor-intensive, and costly, and may be unfeasible in large oilfields with complex multi-well injection parameter combinations. The hunger game search algorithm was introduced to optimize the injection parameters for CO₂ WAG, with the addition of chaotic mapping functions to enhance the randomness and diversity of initial injection parameter values. This new approach formed an improved hunger game search algorithm based on chaotic mapping functions, allowing for collaborative intelligent optimization between the algorithm and reservoir simulation software. This method enhanced the accuracy and efficiency of CO₂ WAG injection parameter optimization for typical oilfields. Compared to the Logistic, Gaussian, and Singer chaotic mapping functions, the Tent chaotic mapping function resulted in more evenly distributed chaotic values and frequency distributions, making it a better choice for improving the hunger game search algorithm. The hunger game search algorithm improved by the Tent chaotic mapping function is an effective method for optimizing CO₂ WAG injection parameters. The optimal CO₂ WAG injection parameters derived from this approach lead to a cumulative oil production of 34.974×10⁴ m³, a 0.213×10⁴ m³ increase over the results from the Hunger Game Search algorithm, and 5.820×10⁴ m³ more than the current CO₂ WAG injection parameter scheme. This approach provides an effective technical solution for the efficient implementation of CO₂ WAG in the field.